| Article Access Statistics | | | Viewed | 2229 | | | Printed | 127 | | | Emailed | 0 | | | PDF Downloaded | 18 | | | Comments | [Add] | | |
|
Click on image for details.
|
|
 |
| LETTER |
|
|
|
| Year : 2014 | Volume
: 60
| Issue : 3 | Page : 344 |
Authors' reply
S Kalra, V Singh
Consultant Endocrinologist, Bharti Hospital, Karnal, Haryana, India
| Date of Web Publication | 14-Aug-2014 |
Correspondence Address:
Dr. S Kalra
Consultant Endocrinologist, Bharti Hospital, Karnal, Haryana
India
 Source of Support: None, Conflict of Interest: None  | Check |
How to cite this article:
Kalra S, Singh V. Authors' reply. J Postgrad Med 2014;60:344 |
Sir,
We thank reviewers and readers [1],[2] for critically analyzing the results of the landmark OBSTACLE Hypoglycemia study. [3] We would like to emphasize once again that linkage between hypoglycemia and glycemic control has been an important challenge in the optimal clinical management of type 2 diabetes. [4] We aimed to demonstrate correlation between these two parameters, in real life clinical setting, using a prospective multi-centric, pan-India study, involving a large number of patients. While the study design does have limitations, concerns raised by Raina SK are not entirely justified. [1] A point by point clarification is provided.When study population was sub-categorized on basis of various parameters like gender, age, body mass index (BMI) and duration of diabetes (years), only those patients were included for whom the relevant information was available. For instance, out of 950 patients considered evaluable, information on gender was not available for 7 subjects, and hence they were excluded from sub-group analysis based on gender. This is evident from the reported data.
Sufficiency of sample size
We do feel that the sample size is sufficient as , the correlation (primary study end-point) has been found to be statistically significant, not just for the entire study population, but also for many of the smaller patient sub-groups [Table 2].
Use of linear regression analysis
As indicated in paragraph 4, of second column, on page 2 of the article, linear regression analysis was indeed the first tool used to assess correlation. However, with data found to be skewed by a majority of hypoglycemia scores being '0', a negative binomial regression model was used to reassess the correlation. [Figure 1] is a graphical representation of negative binomial regression demonstrating weak negative correlation with β slope = -0.09 (95% CI; -0.1468 to 0.0374) and P = 0.0010. Contrary to the claim made by Raina et al., [1] [Figure 1] does not represent linear regression analysis.
Correlation affected by outliers in sub-group analysis by disease duration
When correlation was assessed in patient sub-categories formed on basis of disease duration, it was found to be statistically significant not only in <5 years group that constituted 68.9% (640) of study participants, but also in the 5 years to 10 years group that constituted another 25.7% (239) of study population. Even though the statistical significance of correlation should not be commented upon for >10 years group with numbers of patients being too few (49), the coefficients of correlation was very similar in all the three subgroups (ranging between 0.12 and 0.15). Since the statistical significance of correlation was demonstrated in an overwhelming majority of patients (94.7%), the concept of analysis being impacted by outliers is really not applicable.
| :: Acknowledgments | |  |
This study was sponsored by an unrestricted grant from MSD Pharmaceuticals Pvt. Ltd., India. We would like to thank Sunita Nair and Javed Shaikh, Capita India for editorial support in preparing the manuscript. We would also like to thank Clinigene International Ltd. for clinical data management and analysis.
| :: References | | |
| 1. | Raina SK. Understanding correlation in the context of outliers. J Postgrad Med 2014;60:343-4. 
 |
| 2. | Patell RD, Dosi RV. OBSTACLE hypoglycemia: Targeting a major hurdle in diabetes management! J Postgrad Med 2014;60:161-2.
|
| 3. | Kalra S, Deepak MC, Narang P, Singh V, Maheshwari A. Correlation between measures of hypoglycemia and glycemic improvement in sulfonylurea treated patients with type 2 diabetes in India: Results from the OBSTACLE hypoglycemia study. J Postgrad Med 2014;60:151-5.
[PUBMED] |
| 4. | Ahrén BO. Avoiding hypoglycemia: A key to success for glucose lowering therapy in type 2 diabetes. Vasc Health Risk Manag 2013;9:155-63.
|
|
